IntroductionThe site is situated at the entrance of theYangtzehaven, see figure 1. The length andwidth are 800 m and 200 m respectively. In figure 1b the location is shown in more detail.The height of the slope is 30 m.The reclamation of the site has been carriedout by Boskalis BV. Activities started in August2006, and finished at the end of 2007. Thereclamation was done by means of hydraulicfilling. The slope of the reclaimed site has aheight of 30 m with a steep angle of 1:2.6,determined by nautical requirements. Forstability reasons the slope has been built up bya sequence of bunds consisting of steel slags.Large parts of the filling have been compactedby means of vibroflotation to meet the

safety requirements.This article presents an overview over thedesign, construction, and results. Future articles will present the various aspects in depth.

Soil conditionsOriginally, the location was part of the NorthSea with a bottom level of approximately NAP – 8 m. NAP is the reference level,approximately equal to the mean sea level. After 1960 the Maasvlakte has been reclaimed,until a level of NAP + 5 m. At the projectlocation, no sand filling took place. After completion of the Maasvlake in the south eastcorner of the Yangtzehaven an area was dredgedto a level deeper than NAP – 50 m, to serve as a anchorage berth for offshore equipment.

Furthermore, dredging operations where carriedout for equipment testing. Part of the dredgedmaterial returned in the pits due to overflowfrom the hoppers, and deposited in the deepestareas. This resulted in an unpredictable andrather muddy soil profile.

The traces of this history are still visible in theresults of the soil investigations. First, below the bottom of the Yangtzehaven a mud layerwith a maximum thickness of 2 m was found.Underneath, a loosely packed layer of fine sandwith a maximum thickness of 10 m is present.Further, at some locations a clay layer is presentwith a thickness of 1 up to 3 m. The top level ofthe firm Pleistocene sand varies from NAP – 20 mto NAP – 34 m and locally even more deep.

of a land reclamation

in the

Yangtze Haven in Rotterdam

Abstract

At the entrance of the Yangtzehaven inthe port of Rotterdam a new reclamationhas been carried out. The design of thisreclamation was challenging due to therequirement of sufficient safety againstliquefaction and subsequent slopeinstability, also during the possibleoccurrence of an earthquake.Vibroflotation has been applied to reduceliquefaction risk. Some sand pockets,however, could not be compacted. A sophisticated probabilistic analysis made clear that the realised reclamationmeets the safety requirements.

Leefbaarheid steden Dr.ir. O.M. Heeres1.2 ing. H.E. Brassinga1

ir. J.G. de Gijt1.2 ir. E.J. Broos1 ir. M.B. de Groot3

dr. G.A. van den Ham3 ir. T. Schweckendiek 2.3

62 GEOtechniek Special Deltatechnologie – oktober 2008

4

Geotechnical aspects

Boundary conditionsThe following boundary conditions were crucialfor the design of the reclamation.

- Nautical requirementsThe site is located along the Yangtzehaven,which will be the entrance to the futureMaasvlakte 2. To guarantee an undisturbed ship passage, Rotterdam Port Authority required a slope inclination of 1:2.6.

- Safety requirementThe future user required a failure probability forconstructions on the fill due to slope instabilityand subsequent damage to be smaller than 10-6 during a 50 years design life.

- Earthquake loadingAccording to Eurocode 8 and data from the RoyalDutch Meteorological Institute KNMI (De Crook,1996) the possibility of an earthquake had to beconsidered. Only a small earthquake is possible,with a maximum horizontal acceleration of 0.04 gand an occurrence of once per 10,000 years.

Reclamation designFailure mechanisms and analysisIf no liquefaction is taken into account in the stability analysis, the Bishop stability factor islarger than 1.6. This changes as soon as lique-faction is also taken into consideration. Fromtests performed by Deltares and experience inthe Port of Rotterdam (van Kuijk, 1989; van Rijt,1991; de Gijt et al, 1999) it became clear that the combination of the steep slope and the usedsand can initiate liquefaction inside the lowerpart of the slope, which may lead to instability.

Last, when an instability has occurred, breachingmay cause further slope deterioration(Mastbergen, 2006).

The Deltares methodology (Stoutjesdijk et al.,1998; de Groot et al., 2007) has been appliedto analyse liquefaction. In this method the‘metastable region is computed, in which thecombination of the stresses and the relativedensity is such that liquefaction may occurduring undrained loading’.Major liquefaction risk parameters are theslope geometry, the used material, sand grading,the relative sand density, and the stress state. The earthquake has been accounted for as anadditional load in the slope stability calculations

and was treated in a probabilistic manner byconsidering the full range of accelerations andfailure probability distributions. This approachis similar to the treatment of extreme waterlevels in dike safety calculations (GeoDelft,2007). An earthquake may also serve as trigger.

Stability has been analysed with the conven-tional Bishop method. It should be remarkedthat liquefaction analysis with finite elementsrequires programs in which not only equilibrium,but also the groundwater mass balance isdiscretized. Next, material models must be ableto simulate densification. In terms of elastoplas-ticity this means that the model must be able to produce plasticity within the yield contour.

GEOtechniek Special Deltatechnologie – oktober 2008 63

Figure 1 Site location. Right: top view of the reclamation.

Figure 2 Reduced strength histogram of Monte Carlo results for reclamation sand. Left: liquefiable subset of the outcomes. Middle: non-liquefiable subset. Right: complete set.

Reduced strength histogram

tan(φred) [-] tan(φred) [-] tan(φred) [-]

binc

ount

s [-

]

binc

ount

s [-

]

binc

ount

s [-

]

+ =

During the last decades a host of research hasbeen carried out and it is expected that in thecoming years the first robust programs will beready for use.

MaterialRelative density has been determined from CPTtests. From available relations (Baldi et al.,1982;Lunne et al.,1983; Schmertmann, 1976) Lunne’srelation was chosen as it yields the best resultscomparing to electrical conductivity measure-ments. The uncertainty in the relative densityapplied in the computations consists of twocomponents: an uncertainty of the CPT corre-lation function and an uncertainty due to spatialvariability within the soil body around a CPT.Resulting, the relative density is expressed interms of an expectation and a standarddeviation, serving as input in a probabilisticcomputation.

In order to perform probabilistic stabilityanalyses, the liquefied material shear strengthhas been expressed in terms of a probabilitydistribution function. The reduced friction

angle has been determined as:

tanϕ red =p’min tanϕ, with ϕ red the apparentp’insitu

friction angle in the liquefied situation, p’min

the minimum isotropic effective stress atthe undrained stress path during liquefactionand p’insitu the representative in-situ isotropiceffective stress. The value of p’min is a functionof the relative density, the initial stress and anumber of parameters. Expected values andstandard deviations of these parameters havebeen derived from an extensive series oftriaxial tests. The reduced friction angle hasbeen determined from Monte Carlo analyses.Figure 2 depicts a resulting strength histogram.This histogram can be split into a histogram for liquefiable sand, and non-liquefiable sand,yielding the liquefaction susceptibility.

Table 1 illustrates this for liquefiable sand andcompacted sand. As can be seen, the variationcoefficient = (standard deviation/mean value)is very high in a liquefiable layer. This parameter has a major influence on probabilisticcomputations as large values increase the

failure probability dramatically compared to a deterministic computation.

Measures and resultsThe fill has been constructed in layers with athickness of 2 to 3 m each. Directly at the slopebunds of steel slags have been made. Up to ca.10 m behind these dams a hydraulic fill of coarsesand was applied. At larger distances from thebunds the (medium fine) reclamation sand wasplaced. The latter is known to be more suscep-tible to liquefaction due to its grain distributionand typical relative density in hydraulic fills.

From the liquefaction investigation it turnedout that measures were necessary to fulfill therequired safety requirements. As the applicationof a less steep slope was not possible becauseof nautical reasons, it was decided to increasethe relative density of the fill. For this purposethe ‘vibroflotation method’ was used. As con-struction had already started before the lique-faction aspects were fully recognized, thismethod could not be applied underneath thelowest steel slag dam, which had already beenmade. Also small areas directly behind the steelslag dams could not be reached for compaction. In figure 3 the areas of compaction are indicated(zones 1-4).

The vibroflotation was applied in a triangularpattern of 3.5 m. The required relative densitywas 60% in the sand layers according to Lunne.The result was checked by means of CPT tests.

64 GEOtechniek Special Deltatechnologie – oktober 2008

Figure 3 Zones where vibroflotation was applied. The zone below the first steel slag dam could not be densified.

meanϕ [ ]̊ standard deviation of ϕ [ ]̊ variation coefficient distribution

Liquefiable sand 11.3 5.0 44% Lognormal

Compacted sand 35.9 1.1 3% Normal

Table 1 Shear strength for a liquefiable and compacted reclamation sand.

Geotechniek Special – Deltatechnologie

In Figure 4 the result of densification is illu-strated. Due to their composition, clay, siltysand and clayey sand layers did not or almost not densify (Figure 4). This was confirmed byadditional laboratory tests.The future constructions on the site will beconstructed on a shallow foundation. In orderto limit settlements, during a period of 4 monthsa preload of sand is applied on top of the fillwith a height of 15 m (figure 5).

ConclusionsThe stability design of the embankments washighly influenced by its height and slope, therequired low probability of failure, and the soil conditions.The stability was determined by the possibilityof static liquefaction. The use of steel slags andthe compaction led to a design which fulfilledthe demands.

1 Public Works Rotterdam2 Delft University of Technology3 Deltares

References– Baldi G., Bellotti R., Ghionna V., JamiolkowskiM., Pasqualini E. (1982), Design parameters for sands from CPT, Proc. 2nd European Symposium on Penetration Testing, Amsterdam.– De Crook (1996), A seismic zoning mapconforming to Eurocode 8, and practical earthquake paraeters for the Netherlands,Geologie en Mijnbouw, 75, pp. 11-18.– De Groot, M.B., Stoutjesdijk, T.P., Meijers, P. & Schweckendiek, T. (2007). Verwekings-vloeiing in zand. Geotechniek, oktober 2007, pp 54 - 59.– Gijt J.G. de, Boer L. de, Boxhoorn J.J. (1999), Construction of a sanddam with a height of 50 m in the Europaharbour in Rotterdam, 12th International Harbour Congres, Antwerp, September 6th-12th.– Eurocode 8, part 4 (NVN ENV 1998-4), Design of structures for Earthquake resistance.– GeoDelft (2007), Probabilistic treatment ofearthquake loads, GeoDelft memo CO425460-0062.– GeoDelft (1994), Handboek Zettings-vloeiingen, GD-report CO 353260/10.– Kuyk E.G.J. van, Rijt C. van, Gijt J.G. de (1989), Onderzoek en evaluatie van zandstortenonderwaterdam ECT/Sealand Maasvlakte,Gemeentewerken Rotterdam, november 1989.– Lunne T, Christoffersen H.P. (1983),Interpretation of cone penetrometer data for offshore sands, Proc. of the OffshoreTechnology Conference, Richardson, Texas,paper no. 1464.

– Mastbergen D.R. (2006), Oeverstabiliteit in zandwinputten – rekenmodel HMBreach, Delft Cluster report DC 04 43 11, WL/Hydraulics Z4141.20.– Rijt C. van, Kuijk E.G.J. van, Gijt J.G. de, (1991), Construction of an underwater sanddam for the building excavation for the E.C.T./SealandContainer Terminal on the Maasvlakte, Rotterdam CEDA-PIANC Conference, Accessible Harbour Proceedings.– Schmertmann J.H. (1976), An updatedcorrelation between relative density Dr and Fugro type electric cone q, Contract report. DACW39-76-M6646 WES, Vicksburg.– Stoutjesdijk, T.P,.de Groot, M.B., Lindenberg, J. (1998). Flow slide predictionmethod: influence of slope geometry. Can. Geotechn. J. 35, pp 34. �

GEOtechniek Special Deltatechnologie – oktober 2008 65

Figure 4Left: Vibroflotation.Right: Two CPT's atsmall distance fromeach other. The left CPT is madebefore compaction. The right CPT is made after compaction.The effect on the sandlayers is clearly visible. The line of 60% relative density is depicted for reclamation sand (Lunne et al., 1983).Photo: Keller